This is an archive of the discontinued LLVM Phabricator instance.

Differential D77688

[CUDA] Improve testing of libdevice detection.
ClosedPublic

Authored by tra on Apr 7 2020, 3:55 PM.

Details

Reviewers
yaxunl
Commits
rGd2e498b1725d: [CUDA] Improve testing of libdevice detection.
Summary

Added new testcases for libdevice in CUDA-9+ and removed unused checks.

Diff Detail

Event Timeline

tra created this revision.Apr 7 2020, 3:55 PM
Herald added a project: Restricted Project. · View Herald TranscriptApr 7 2020, 3:55 PM
Herald added subscribers: sanjoy.google, bixia. · View Herald Transcript
Harbormaster failed remote builds in B52258: Diff 255845!Apr 7 2020, 4:55 PM
yaxunl accepted this revision.Apr 7 2020, 8:48 PM

LGTM. Thanks.

This revision is now accepted and ready to land.Apr 7 2020, 8:48 PM
Closed by commit rGd2e498b1725d: [CUDA] Improve testing of libdevice detection. (authored by tra). · Explain WhyApr 8 2020, 11:24 AM
This revision was automatically updated to reflect the committed changes.

Revision Contents

PathSize
clang/
test/
Driver/
43 lines

Diff 255845

clang/test/Driver/cuda-detect.cu

Show All 37 Lines
// RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE // RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE
// RUN: %clang -v --target=i386-apple-macosx -nocudalib \ // RUN: %clang -v --target=i386-apple-macosx -nocudalib \
// RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE // RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE
// RUN: %clang -v --target=x86_64-unknown-linux -nocudalib \ // RUN: %clang -v --target=x86_64-unknown-linux -nocudalib \
// RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE // RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE
// RUN: %clang -v --target=x86_64-apple-macosx -nocudalib \ // RUN: %clang -v --target=x86_64-apple-macosx -nocudalib \
// RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE // RUN: --sysroot=%S/Inputs/CUDA-nolibdevice --cuda-path-ignore-env 2>&1 | FileCheck %s -check-prefix NO-LIBDEVICE
Lint: Pre-merge checks
Inline Actions

clang-format: please reformat the code

-
Lint: Pre-merge checks: clang-format: please reformat the code ``` - ```
// Make sure we map libdevice bitcode files to proper GPUs. These // Make sure we map libdevice bitcode files to proper GPUs. These
// tests use Inputs/CUDA_80 which has full set of libdevice files. // tests use Inputs/CUDA_80 which has full set of libdevice files.
// However, libdevice mapping only matches CUDA-7.x at the moment. // However, libdevice mapping only matches CUDA-7.x at the moment.
// sm_2x, sm_32 -> compute_20 // sm_2x, sm_32 -> compute_20
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_21 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_21 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON \ // RUN: | FileCheck %s -check-prefix COMMON \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE20 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE20
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_32 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_32 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON \ // RUN: | FileCheck %s -check-prefix COMMON \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE20 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE20
// sm_30, sm_6x map to compute_30. // sm_30, sm_6x map to compute_30.
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_30 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_30 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON \ // RUN: | FileCheck %s -check-prefix COMMON \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE30 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE30
// sm_5x is a special case. Maps to compute_30 for cuda-7.x only. // sm_5x is a special case. Maps to compute_30 for cuda-7.x only.
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_50 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_50 \
// RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON \ // RUN: | FileCheck %s -check-prefix COMMON \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE30 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE30
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_60 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_60 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON \ // RUN: | FileCheck %s -check-prefix COMMON \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE30 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE30
// sm_35 and sm_37 -> compute_35 // sm_35 and sm_37 -> compute_35
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix CUDAINC \ // RUN: | FileCheck %s -check-prefix COMMON -check-prefix CUDAINC \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE35 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE35
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_37 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_37 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix CUDAINC \ // RUN: | FileCheck %s -check-prefix COMMON -check-prefix CUDAINC \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE35 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE35
// sm_5x -> compute_50 for CUDA-8.0 and newer. // sm_5x -> compute_50 for CUDA-8.0 and newer.
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_50 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_50 \
// RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \ // RUN: --cuda-path=%S/Inputs/CUDA_80/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON \ // RUN: | FileCheck %s -check-prefix COMMON \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE50 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE50
// CUDA-9+ uses the same libdevice for all GPU variants:
// RUN: %clang -### -v --target=x86_64-unknown-linux --cuda-gpu-arch=sm_30 \
// RUN: --cuda-path=%S/Inputs/CUDA_90/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON64 \
// RUN: -check-prefixes PTX60,LIBDEVICE,LIBDEVICE10
// RUN: %clang -### -v --target=x86_64-unknown-linux --cuda-gpu-arch=sm_50 \
// RUN: --cuda-path=%S/Inputs/CUDA_90/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON64 \
// RUN: -check-prefixes PTX60,LIBDEVICE,LIBDEVICE10
// RUN: %clang -### -v --target=x86_64-unknown-linux --cuda-gpu-arch=sm_60 \
// RUN: --cuda-path=%S/Inputs/CUDA_90/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON64 \
// RUN: -check-prefixes PTX60,LIBDEVICE,LIBDEVICE10
Lint: Pre-merge checks
Inline Actions

clang-format: please reformat the code

-
Lint: Pre-merge checks: clang-format: please reformat the code ``` - ```
// Verify that -nocudainc prevents adding include path to CUDA headers. // Verify that -nocudainc prevents adding include path to CUDA headers.
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \
// RUN: -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \ // RUN: -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC \ // RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE35 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE35
// RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_35 \ // RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_35 \
// RUN: -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \ // RUN: -nocudainc --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC \ // RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC \
// RUN: -check-prefix LIBDEVICE -check-prefix LIBDEVICE35 // RUN: -check-prefixes PTX42,LIBDEVICE,LIBDEVICE35
// We should not add any CUDA include paths if there's no valid CUDA installation // We should not add any CUDA include paths if there's no valid CUDA installation
// RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \ // RUN: %clang -### -v --target=i386-unknown-linux --cuda-gpu-arch=sm_35 \
// RUN: --cuda-path=%S/no-cuda-there %s 2>&1 \ // RUN: --cuda-path=%S/no-cuda-there %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC // RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC
// RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_35 \ // RUN: %clang -### -v --target=i386-apple-macosx --cuda-gpu-arch=sm_35 \
// RUN: --cuda-path=%S/no-cuda-there %s 2>&1 \ // RUN: --cuda-path=%S/no-cuda-there %s 2>&1 \
// RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC // RUN: | FileCheck %s -check-prefix COMMON -check-prefix NOCUDAINC
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK: Found CUDA installation: {{.*}}/Inputs/CUDA/usr/local/cuda // CHECK: Found CUDA installation: {{.*}}/Inputs/CUDA/usr/local/cuda
// NO-LIBDEVICE: Found CUDA installation: {{.*}}/Inputs/CUDA-nolibdevice/usr/local/cuda // NO-LIBDEVICE: Found CUDA installation: {{.*}}/Inputs/CUDA-nolibdevice/usr/local/cuda
// NOCUDA-NOT: Found CUDA installation: // NOCUDA-NOT: Found CUDA installation:
// MISSINGLIBDEVICE: error: cannot find libdevice for sm_20. // MISSINGLIBDEVICE: error: cannot find libdevice for sm_20.
// COMMON: "-triple" "nvptx-nvidia-cuda" // COMMON: "-triple" "nvptx-nvidia-cuda"
// COMMON64: "-triple" "nvptx64-nvidia-cuda"
// COMMON-SAME: "-fcuda-is-device" // COMMON-SAME: "-fcuda-is-device"
// LIBDEVICE-SAME: "-mlink-builtin-bitcode" // LIBDEVICE-SAME: "-mlink-builtin-bitcode"
// NOLIBDEVICE-NOT: "-mlink-builtin-bitcode" // LIBDEVICE10-SAME: libdevice.10.bc
// LIBDEVICE20-SAME: libdevice.compute_20.10.bc // LIBDEVICE20-SAME: libdevice.compute_20.10.bc
// LIBDEVICE30-SAME: libdevice.compute_30.10.bc // LIBDEVICE30-SAME: libdevice.compute_30.10.bc
// LIBDEVICE35-SAME: libdevice.compute_35.10.bc // LIBDEVICE35-SAME: libdevice.compute_35.10.bc
// LIBDEVICE50-SAME: libdevice.compute_50.10.bc // LIBDEVICE50-SAME: libdevice.compute_50.10.bc
// NOLIBDEVICE-NOT: libdevice.compute_{{.*}}.bc // PTX42-SAME: "-target-feature" "+ptx42"
// LIBDEVICE-SAME: "-target-feature" "+ptx42" // PTX60-SAME: "-target-feature" "+ptx60"
// NOLIBDEVICE-NOT: "-target-feature" "+ptx42"
// CUDAINC-SAME: "-internal-isystem" "{{.*}}/Inputs/CUDA{{[_0-9]+}}/usr/local/cuda/include" // CUDAINC-SAME: "-internal-isystem" "{{.*}}/Inputs/CUDA{{[_0-9]+}}/usr/local/cuda/include"
// NOCUDAINC-NOT: "-internal-isystem" "{{.*}}/cuda/include" // NOCUDAINC-NOT: "-internal-isystem" "{{.*}}/cuda/include"
// CUDAINC-SAME: "-include" "__clang_cuda_runtime_wrapper.h" // CUDAINC-SAME: "-include" "__clang_cuda_runtime_wrapper.h"
// NOCUDAINC-NOT: "-include" "__clang_cuda_runtime_wrapper.h" // NOCUDAINC-NOT: "-include" "__clang_cuda_runtime_wrapper.h"
// -internal-externc-isystem flags must come *after* the cuda include flags, // -internal-externc-isystem flags must come *after* the cuda include flags,
// because we must search the cuda include directory first. // because we must search the cuda include directory first.
// CUDAINC-SAME: "-internal-externc-isystem" // CUDAINC-SAME: "-internal-externc-isystem"
// COMMON-SAME: "-x" "cuda" // COMMON-SAME: "-x" "cuda"
Show All 17 Lines